ES2719608T3 - Improvements in the determination of hierarchies for energy monitoring systems - Google Patents

Abstract

A machine learning procedure of how multiple monitoring devices (M1 to M13) are directly or indirectly linked in a monitoring system, in which the monitoring system uses an Ethernet (704) and comprises at least one Ethernet gateway (700, 701), comprising the procedure: determining an IP address for each of said monitoring devices (M1 to M13), in which the monitoring devices connected to the Ethernet through the same Ethernet gateway (700, 701) have the same IP address; receiving voltages measured by each of said monitoring devices; grouping said monitoring devices into multiple segments based on the determined IP addresses, wherein the monitoring devices having the same IP address are grouped into one segment; and determining, within the segments, potential relationships between the monitoring devices based on said received measured voltages.

Description

DESCRIPTION

Improvements in the determination of hierarchies for energy monitoring systems

^{FIELD OF THE INVENTION}

The present invention generally relates to supply monitoring systems and, in particular, to the automated alignment of data accuracy, the automated determination of the hierarchy of an energy monitoring system, and the automated integration of data into a delivery system. Supply monitoring

BACKGROUND

Since the introduction of electric power distribution systems at the end of the 19th century, it has been necessary to monitor their operational and electrical characteristics. The ability to collect, analyze and respond to information about the electric power system can improve safety, minimize equipment loss, decrease excess energy and, ultimately, save time and money. To this end, monitoring devices were developed to measure and report such information. With the beginning of the era of electronics, the quality and quantity of data from monitoring devices greatly improved, and communications networks and software were developed to collect, display and store information. Unfortunately, those responsible for evaluating the data of the monitoring devices are now overwhelmed by the information of their monitoring systems. In the effort to maximize the public service of a monitoring system, manufacturers of monitoring equipment are looking for procedures to present information in the most useful format.

Effective monitoring of today's electricity distribution systems is cumbersome, expensive and inefficient. Electric power monitoring systems are typically arranged in a hierarchy with monitoring devices such as electrical meters installed at various levels of the hierarchy (see Figure 2). The monitoring devices measure various characteristics of the electrical signal (for example, voltage, current, waveform distortion, power, etc.) that passes through the conductors, and the data from each monitoring device is analyzed by the user to evaluate the potential performance or quality related problems. However, the components of current electrical monitoring systems (monitoring devices, software, etc.) act independently of each other, which requires the user to be an expert in hardware configuration, collection and analysis of data and the determination of what data is vital or useful. There are two problems in this case: the amount of data to analyze and the context of the data. These are separate but related issues. It is possible to automate data analysis to address the amount of data. But, in order to do this reliably, the data must be put into context. The independence of the data between each monitoring device that evaluates the electrical system essentially makes each monitoring device foreign to the data coming from other monitoring devices connected to the system being analyzed. Accordingly, the data transmitted to the computer device 132 of the system from each monitoring device is usually misaligned, since the data from each system monitoring device does not reach the computer device 132 of the monitoring system simultaneously. There are two basic reasons for the temporary misalignment of the data between the monitoring devices: the time delays of the communications and the timing of the monitoring devices and the time recording of the events. Next, it is up to the user to analyze and interpret these independent data in order to optimize performance or evaluate possible concerns related to the quality of the electrical system.

The sophisticated processing capabilities of digital monitoring devices make it possible to derive and accumulate large amounts of complex electrical data from a seemingly simple electrical signal. Due to the complexity of the data, the quantity and the relatively disarticulated relationship from one monitoring device to the next, the manual analysis of all the data is a huge effort that often requires the hiring of experts to complete the task. This process is tedious, complex, prone to mistakes and carelessness, and time consuming. A partial solution has been to use global satellite positioning (GPS) systems to record the time of an event, but this approach requires the user to purchase and install additional hardware and data lines to connect the monitoring devices to each other. And this solution still requires the evaluation of large amounts of data because the system is temporarily in context; not spatially in context. Data synchronization using GPS systems is also disadvantageous due to time delays associated with other hardware in the system. Loss of the signal from GPS satellites makes this approach not work. In addition, any alignment of data using a GPS-based system can only be as accurate as the delay in propagation of the GPS signal, which means that the data may not be optimally aligned when a GPS system is used.

The aggregation of supplementary monitoring devices in the electrical system only generates more information about the electrical system at the point where the meter is added to the electrical system, increasing complexity without any benefit. Any utility of the data is generally limited to the location of the monitoring device that has been added, while further data is accumulated.

The complexity of many electrical systems normally requires a process of configuring monitoring systems because each point measured in the electrical system has different characteristics, which is why multiple monitoring devices are installed first. As a result of the enormous volume of complex data accumulated from the electrical monitoring systems so far, a thorough analysis of the data is usually not feasible due to the limitation of resources, time and / or experience.

The temporal alignment of the data is an important aspect to understand and characterize the electrical system. Another important aspect is to have a thorough knowledge of the arrangement (or hierarchy) of the energy monitoring system. The energy monitoring devices measure the operational parameters of the electrical system, but do not provide information on how the parameters are related to each other at different points of the energy monitoring system. Knowing the hierarchy of the energy monitoring system puts the operating parameters of multiple monitoring devices in context with each other.

To determine the layout of an energy monitoring system, the user must review the single power line drawings or physically perform an inventory of the electrical system if no single line drawings are available. The user manually enters the spatial information in the monitoring system software for analysis. When a new device or monitored load is added or moved within the energy monitoring system, the user must manually update the monitoring system software to reflect the new aggregation or change.

Data alignment and arrangement information are essential to understand and characterize the energy system. With these two information, the data of each meter can be integrated and put into context with each of the other meters in the energy system. Until now, the only techniques for integrating data in an acceptable way were complex, expensive, manual and intensive, and time consuming for the user. These techniques also allow only limited data integration and require additional hardware (such as GPS hardware), data lines and supplementary accessories for monitoring devices.

Document US 2007/0005275 A1 refers to an automated integration of data into supply monitoring systems. An automated integrated monitoring algorithm (IM) is described that automatically puts the data from a supply monitoring system into context by temporarily aligning it with a common reference point and identifying the location of each monitoring device in a hierarchy in relation to others. dispositives.

WO 2007/027086 A1 refers to techniques for determining the network structure or arrangement of at least a part of an electricity transport network, for example, a low voltage part of this network that extends into a city, which comprises: providing a test signal to the network at least in a first location of the network and verification of whether said test signal, which has been provided to the first location of the network, reaches at least a second location of the network, in particular using at least one detector.

SUMMARY OF THE INVENTION

The object of the present invention is to improve the prior art systems. This object is solved by the subject of the independent claims. The dependent claims define preferred embodiments.

The foregoing and additional aspects of the present invention will be apparent to those skilled in the art in view of the detailed description of various embodiments, which makes reference to the drawings, and a brief description thereof is given below.

BRIEF DESCRIPTION OF THE DRAWINGS

The above advantages and other advantages of the invention will be apparent upon reading the following detailed description and with reference to the drawings.

Figure 1 is a functional block diagram of an automated data integration monitoring system;

Figure 2 is a functional block diagram of a simplified energy monitoring system;

Figure 3 is a functional block diagram of a monitoring device;

Figure 4 are samples of sample frequency data of two aligned monitoring devices;

Figure 5A is a flow chart of a data alignment algorithm;

Figure 5B is a flow chart of a modified data alignment algorithm;

Figure 6 is a functional block diagram of a simplified hierarchy with a single main feeder and two feeders;

Figure 7 is an example diagram of a simple radial feed system;

Figure 8 is an example diagram of a multiple radial feed system;

Figures 9-11A is a flow chart of a machine learning hierarchy algorithm;

Figure 11B is a flow chart of a machine learning hierarchy algorithm;

Figure 11C is a flow chart of a modified machine learning hierarchy algorithm;

Figure 12 is a flow chart of an automated integrated monitoring algorithm;

Figure 13 is a diagrammatic illustration of an example segmentation of an energy monitoring system;

Figure 14 is a diagrammatic illustration of an energy monitoring system that includes an Ethernet, a pair of Ethernet gateways, and multiple energy monitoring devices;

Figure 15 is a diagrammatic illustration of a logical segmentation of the devices included in the system of Figure 14 and includes potential relationships of some of the devices based on voltages measured by the devices;

Figure 16 is a diagrammatic illustration of hierarchies of the devices in each of the segments illustrated in Figure 15;

Figure 17 is a hierarchical representation of an energy monitoring system comprising the most advanced devices in all segments illustrated in Figure 15;

Figure 18 is a hierarchical representation of an energy monitoring system comprising all devices in the segments illustrated in Figure 15;

Figure 19 is a flow chart of a segmentation process of an energy monitoring system; Figure 20 is a representation of a part of an energy monitoring system indicating an energy consumption greater than the energy delivered;

Figure 21 is a representation of a part of an energy monitoring system indicating a lower energy consumption than the energy delivered, with a virtual VM meter representing an unmonitored load; Figure 22 is a representation of a part of an energy monitoring system indicating a lower energy consumption than the energy delivered;

Figure 23 is a sample of a 24-hour charging profile of a customer;

Figure 24 illustrates an example of adaptive sampling for the load profile of Figure 26;

Figure 25 is a flow chart of a process that implements improved techniques for determining and evaluating hierarchies;

Figure 26 is a flow chart of a process looking for the children of a reference device;

Figure 27 is a diagrammatic illustration of a part of a hierarchy of a monitoring system that includes an orphan device represented by a virtual monitoring device; Y

Figure 28 is a diagrammatic illustration of the hierarchical part shown in Figure 27 with the orphan device placed in the hierarchy.

While the invention is susceptible to various modifications and alternative forms, specific embodiments have been shown by way of example in the drawings and will be described in detail herein. It should be understood, however, that the invention is not intended to be limited to the particular forms disclosed. Rather, the invention covers all modifications, equivalents and alternatives that fall within the spirit and scope of the invention as defined by the appended claims.

DETAILED DESCRIPTION

Referring now to Fig. 1, an automated integrated data monitoring system 100 is generally shown. A delivery system 102 having multiple monitoring devices M provides data of each monitoring device M that communicates with an alignment system of Automated data 104 and with an automated hierarchical classification system 106. The data is automatically aligned in the automated data alignment system 104 in accordance with the present invention and produces data that is aligned such that they represent the data that was actually observed. simultaneously by the monitoring devices M in the energy monitoring system 102. The hierarchical classification system 106 automatically learns the hierarchy of monitoring devices that are present in the supply system 102 and their relationships with each other.

Once the data of each monitoring device M is aligned and the location of each monitoring device is known, it is understood that the data is in context 108. The contextual data 108 can be used by the software applications 110 to provide and diagnose useful information about the delivery system 102 beyond what is generally available if the data is not in context. The public service that is monitored in system 102 can be any of the five public services designated by the acronym, WAGES (water, air, gas, electricity, or steam), or water, air, gas, electricity or steam. Each monitoring device measures characteristics of the delivery system and quantifies these characteristics in data that can be analyzed by a computer device.

A user interacts with software applications 110 through a conventional user interface 112. Software applications 110 may be linked to other systems 114, such as a billing system, and use contextual data 108 to communicate messages between the other systems 114 and user interface 112.

The data alignment system 104 aligns data, such as voltage / voltage, current, time, events and the like, of multiple monitoring devices M in a delivery system, and is a valuable tool for users. When the data from all monitoring devices M is aligned with the same point in time at which the data was produced, they can be placed in a temporary context from which additional decisions on the data can be made or automatically recommended. Hardware and software configuration. As used herein, a monitoring device refers to any element or apparatus of the system with the ability to sample, collect or measure one or more features or operational parameters of a delivery system 102. When the system 102 is An energy monitoring system, the monitoring device M can be a meter that measures electrical characteristics or parameters of the energy monitoring system.

The data alignment techniques (detailed below) achieve at least the following:

1) Automated data alignment in monitoring devices;

2) Automatic time synchronization on monitoring devices;

3) Alignment of data and time in monitoring devices located in different points of the electricity network (in which the monitoring system software can obtain time data from the Internet or another server); and 4) Diagnose misidentification or incorrect phase labeling in the entire electrical power system. In the real world, all electrical signals in energy systems experience subtle variations in their frequency and amplitude over time. This variation in the frequency and amplitude of the signal is indeterminate and unique with respect to time. Each monitoring device located in the same supply network will experience the same frequency variations simultaneously. The analysis of data from monitoring devices that are directly linked to each other in the hierarchy will reveal a correlation in their amplitude variations. The analysis of the variations in frequency and amplitude of the signal is used to accurately align the data of a monitoring device with respect to another device (or all monitoring devices with each other) in the data alignment system 104. The Details of the data alignment system 104 are discussed below.

Data alignment techniques allow all M monitoring devices to be aligned in a hierarchy of power supply system with zero crossing of the three phase voltages without the use of additional hardware. The system also anticipates possible phase shifts between various monitoring devices, for example, those caused by certain transformer configurations. Once the data from the monitoring devices are aligned with each other, the system data is essentially aligned with respect to the moment in which they were produced, which makes it possible to perform more complex data analysis. A simplified configuration of an energy monitoring system 120 is shown in Figure 2. The energy monitoring system 120 includes a main 122 connected to a first load 124 by a first feeder and a second load 126 by a second feeder. The monitoring devices 128, 130 measure electrical characteristics or parameters associated with the first and second feeder, respectively. Each monitoring device 128, 130 is communicatively coupled to a computing device 132. The first monitoring device 128 may be an energy meter (or an electric meter), as shown in Figure 3. The monitoring device 128 includes a controller 134, firmware 136, a memory 138, a communications interface 140 and three-phase voltage conductor connectors 142a, b, c, which are connected to the phase voltage conductors V ^a , V ^b and V ^c , respectively, and are coupled to controller 134. The three-phase current conductor connectors 143a, b, c, which are connected to phase current conductors I ^a , I ^b and I ^c , respectively, are optionally coupled to controller 134 Firmware 136 includes machine instructions to make the controller perform the necessary operations for the monitoring device. Memory 138 is used by controller 134 to store electrical parameter data measured by monitoring device 128.

The monitoring device 128 receives instructions from the computing device 132 through the communications interface 140. These instructions include instructions that cause the controller 134 to dial the cycle count, to begin storing electrical parameter data, or transmitting to the software of the monitoring system 132 electrical parameter data stored in memory 138. The electrical parameter data may include any data acquired by the monitoring devices, including any combination of frequency variations, amplitude variations and phase variations.

An algorithm is used to accurately, automatically and temporally align the data from multiple monitoring devices with respect to the same voltage zero crossing. Other aspects of data alignment discussed below are based on this capability. The alignment of data is facilitated by the functionality in both the monitoring device 128 and the monitoring system software running in the computing device 132, and the requirements of each of them will be discussed individually. The collection and partial analysis of data are carried out in the monitoring device 128.

From the moment that the monitoring device 128 is supplied with energy, a cycle count of the measured voltage signals is performed. Cycle counting is iterated sequentially with each zero crossing of positive voltage (or, alternatively, each zero crossing of negative voltage). As the monitoring device 128 measures the variations in frequency and amplitude of the voltage and current from one cycle to another, a comparison is made with their respective nominal values. Variations in frequency and amplitude and associated cycle count are monitored by the firmware of device 136. The time / time of the associated monitoring device in any specified cycle count can be stored in memory 138.

The monitoring system software executed by the computing device 132 initiates the alignment of the data associated with multiple monitoring devices by sending a global command to all the monitoring devices 128, 130 of the energy monitoring system 120 to mark the cycle count , time and buffer a predetermined amount of data cycle by cycle.

This predetermined amount of data is established based on the number of monitoring devices in the energy monitoring system, the delays in the communication time in the energy monitoring system and the magnitude of the frequency and amplitude variations. Once the buffering is finished, the monitoring devices 128, 130 transmit their buffered data to the computing device 132.

Once the data is collected by the 128,130 monitoring devices, the monitoring system software uploads / loads the stored data for analysis. It is likely that there is a time lag in the data stored in the buffer of each monitoring device because it is likely that the system monitoring devices do not begin to buffer the data simultaneously due to delays in the communication time in the Energy monitoring system and internal time delays within the monitoring devices. The data stored in the buffer is analyzed by the monitoring system software in the computing device 132 to locate the highest frequency correlation among all monitoring devices 128, 130. Generally, the highest correlation is located by sliding the stored frequency data in the buffer of one monitoring device with respect to another until the frequency variations are aligned with each other as shown in Figure 4.

The frequency data 360 for the monitoring device 128 is "slipped" in relation to the frequency data 362 for the monitoring device 130 until the frequency data for each device is aligned. Thus, the zero crossing associated with At ¹ of the monitoring device 128 is aligned with the zero crossing associated with At ¹ of the monitoring device 130, the zero crossing associated with At ² of the monitoring device 128 is aligned with the crossing by zero associated with At ² of the monitoring device 130, and so on. The cross-correlation algorithms for the relative "slippage" of two sets of data with each other until they are aligned are discussed in more detail below in relation to Figures 5A and 5B. Once the data stored in buffer is aligned, the cycle count of the first monitoring device 128 is associated with the cycle count of the second monitoring device 130 in the software of the computing device 132. Optionally, the time of the monitoring device On board can also be aligned or associated with each other. This process is repeated for each monitoring device of the energy monitoring system 120 until all the device cycle counts are associated with each other. During the data alignment process, the software of the monitoring system of the computing device 132 constructs an array of the cycle and time counting of each device with each other and of the time in the computing device 132.

Although Figure 2 shows a simplified energy monitoring system 120 with only two monitoring devices 128, 130, the data alignment can be applied to any energy monitoring system 120 of any complexity with multiple hierarchical levels, such as the diagram single line shown in Figure 7. To facilitate illustration and discussion, only two monitoring devices 128, 130 have been discussed.

Once the data of the two monitoring devices 128, 130 are aligned with each other, there is normally no need to realign the data again unless a monitoring device loses its signal. voltage that is reset by itself. In these cases, it is only necessary to realign the monitoring devices that lose their voltage signal or are reset. The data alignment technique can be initiated by an event, such as an over-voltage or under-voltage condition, the connection or disconnection of a load to the energy monitoring system, a change in the characteristics of the voltage, current, or a load, a reset of a monitoring device or a loss of energy. The data alignment technique can also be initiated automatically by the monitoring software or manually by the user.

Turning now to Figure 5A, a flow chart is shown, which can be implemented as a data alignment algorithm 180 executed by the computing device 132, to perform a data alignment algorithm 180 that begins by sending a message to the devices. monitoring (such as monitoring devices 128, 130) to start buffering data (200) until buffering (202) is completed. The computing device 132 reads the data from each device (204). The data represent, in one aspect, data of electrical parameters such as variations in the frequency (fundamental), variations in the amplitude and variations in the phase. Preferably, the data represents variations in the fundamental frequency. The fundamental frequency is a preferred criterion because it remains unchanged throughout the energy monitoring system, even if there are transformers present in the system. The amplitude and phases may vary when transformers are present in the system; however, the system described in this document contemplates the use of amplitude and phase information as criteria.

The computing device 132 selects a reference monitoring device (206) such as the monitoring device 128 and then selects a monitoring device to analyze (208) such as the monitoring device 130. Then, the data from the monitoring devices Monitoring 128, 130 are correlated with each other (210), and the cycle count and time relationships of each device are included in a matrix (212). Cross-correlation is performed by a conventional cross-correlation algorithm, preferably as provided in Equation 1 below.

The correlation coefficient is represented by r (d), the delay (offset or offset) being represented by d, in which -1 <= r (d) <= 1 for two series x (i) ey (i) that they represent the respective data from the monitoring devices 128, 130; and mx and my are the means of the corresponding series x (i) and y (i). According to one aspect, the correlation algorithm is a circular correlation algorithm in which out-of-range indexes are "wrapped" back into the range. In another aspect, the correlation algorithm is a linear correlation algorithm in which each series is repeated. In other aspects, the correlation algorithm is a pattern matching algorithm or a text search algorithm.

After cross correlation, the computer device 132 checks if all the monitoring devices (214) have been analyzed and, if so, proceeds to check the wiring of the phase conductors. In many cases, phase conductors may be misidentified in an electrical system by the contractor that installed them. For example, the phase that is identified as "phase A" in the main switching equipment may be identified as "phase B" in the load. This misidentification of the nomenclature of the phase conductors can lead to confusion, and even pose a safety hazard.

To mitigate this danger, the computing device 132 analyzes the voltage (or current) data by sampling data at the zero crossing voltage (or current) of a reference channel in each monitoring device (216). The computing device 132 determines if the wiring is correct (218) by determining whether the values of the sampled data are zero, negative or positive and, based on those values, assigning phase notations (such as A, B or C) for each reference channel If all monitoring devices are accurately identified, the data values of phase A should be approximately zero. If the data values are negative, then the phase in question is "Phase B" for an ABC phase rotation. If the data values are positive, then the phase in question is the "phase C" for an ABC phase rotation. The user is notified (220) about whether the wiring is correct. Once the appropriate phase notation for each monitoring device (222) has been determined, the computing device 132 may allow the user to correct the misidentified phase notation in any or all of the monitoring devices. The examples of phase diagnostics are applicable to both voltage inputs and current inputs.

Figure 5B illustrates a flowchart for performing a process in which a computing device 132 supplies instructions to each monitoring device of the energy monitoring system 120 to store data on a cycle-by-cycle basis (250) by a number predetermined cycles, preferably between around 1,000 and around 10,000 cycles. When the monitoring devices have stored a sufficient amount of data, the computing device 132 receives the data from the monitoring devices (252) and selects a reference monitoring device (254). Using a conventional cross-correlation algorithm such as that in equation 1 above, the computing device 132 calculates a correlation coefficient r (d) between at least a part of the data (for example, about 400 cycles) of the reference monitoring device and the data of a second monitoring device (256). The calculated correlation coefficient is stored, and the data of the second monitoring device is shifted in a cycle with respect to the reference device (258).

As mentioned earlier, out-of-range indices can be wrapped again within the range according to a circular correlation algorithm or the indices can be repeated according to a linear correlation algorithm. A correlation coefficient is calculated using the displaced data (260) and, if no further displacements (262) are required, the data of the second monitoring device is aligned with the data of the reference device at the point at which the maximum correlation coefficient or in which the correlation coefficient exceeds a threshold value, such as 0.5 (264). It should be taken into account that when the correlation coefficient r (d) is close to 1.0, the algorithm can end without further displacements.

The computing device 132 synchronizes the clocks of the second monitoring device and the reference device at the alignment point (266). The computing device 132 reads the cycle count of each monitoring device and the time of the on-board clock of the associated monitoring device. The time of the on-board clock and the cycle count of a monitoring device may deviate from each other due to the limitations of the on-board clock. Once the data is aligned, the cycle count is considered the absolute reference for a monitoring device. Due to the deviation of the clock, it may be necessary to re-read the time associated with the counting of cycles of a device periodically to reset the time / time of the device. The software of the computing device 132 will then update the matrix containing the time information of the monitoring device.

Another capability of this function is to allow all clocks of on-board monitoring devices to be periodically reset to the same value to provide a standard time for the entire energy monitoring system. Preferably, the time / time within the monitoring system software (running on the computing device) is adjusted according to an absolute time reference. After setting the time of the computing device, the monitoring system software resets the time of all monitoring devices accordingly. In this example, the data and time of each monitoring device and the software would align more precisely with the absolute time / time reference.

When there are no more monitoring devices to align (268), the procedure ends. In an alternative example, all the data of the monitoring device is aligned before the clocks are synchronized (266).

Another advantage of data alignment techniques is the ability to align data and time at different points in the supply network. If the monitoring devices are located at two different points of the same power supply network, it is possible to align the monitoring devices with each other. In this example, the monitoring devices are aligned first at each geographic location with each other. The software that manages all systems is used as an absolute time reference for all systems, which provides them with a common reference point.

Referring again to Figure 1, the integrated monitoring system 100 includes the hierarchical classification system 106. Having a thorough knowledge of the arrangement of an electrical energy system is essential to understand and characterize the system. Energy meters normally only provide the operational parameters of the electrical system, but do not provide information on how the parameters relate to each other at the different control points of the electrical system. Having the hierarchy of an electrical system places the operational parameters of multiple monitoring devices in a spatial context with each other. This spatial context gives the user a more powerful tool to solve system problems, improve system efficiency, predict failures and degradation, locate the source of disturbances, or model system responses.

The hierarchical classification system 106 allows the monitoring system software to collect data from the monitoring system of the supply system 102, and automatically determine the hierarchy of the supply system 102 with little or no user input. The level of detail given by the hierarchical classification system 106 is directly correlated with the number and scope of the monitoring devices in the delivery system 102. As supplementary monitoring devices are added, the hierarchical machine learning algorithm allows that are automatically incorporated into the determined hierarchical structure.

A hierarchy of nodes is based on a relationship that determines that a node is always larger than another node, when the nodes are related. The relationship of a hierarchy can link or interrelate elements in one of three ways: direct, indirect or unrelated. Figure 6 shows an illustration of a direct relationship or interrelation between a Load ² 310 and a Power Supply ^ 306. On the contrary, there is an indirect link between Load ² 310 and Principan 302. Finally, there is no effective connection between Load ¹ 308 and Load ² 310 or between the Powered ^ 304 and the Powered ^ 306.

In the case of an energy system hierarchy, the objective is to order the elements of the energy system to represent the true connection scheme of the energy system. Determining the hierarchy of an electrical system provides important information that can be used to solve problems, increase equipment and system performance, improve security and save money. The level of detail contained in the hierarchy of an energy system will depend both on the number of elements or nodes that are being monitored and the ability of the node to provide feedback to the machine learning hierarchy algorithm in the monitoring system software that is runs on the computing device 132. Generally, the hierarchical classification system 106 uses a machine learning hierarchy algorithm in the monitoring system software that is based on statistical rules and procedures. Periodically, the monitoring system software polls each monitoring device in the delivery system 102 to determine certain characteristics or parameters of the delivery system 102 in that node (represented by the monitoring device M). Multiple samples of specified parameters are taken from each system meter at the same point in time. Once the parameter data of each node M has been collected in the delivery system 102, the machine learning hierarchy algorithm analyzes the data and tracks the relationships or links between the monitoring devices with respect to the moment at which it was taken the data sample and the associated value of the data sample. This analysis can be performed periodically to increase the probability that the hierarchy is accurate, or to determine any changes in the hierarchy. Once this iterative process reaches a predetermined level of statistical confidence that the determined provision of the delivery system 102 is correct, the machine learning hierarchy algorithm ends. The final provision of the delivery system 102 is presented to the user for approval. As each of the data of each monitoring device is evaluated over time (the learning period) with respect to all other monitoring devices that use the machine learning hierarchy algorithm, a basic arrangement is determined of the hierarchical structure of the supply system 102 based on the available monitoring points. In this sense, the algorithm uses the historical trends of the data from each monitoring device, and these trends are compared to determine if there is any interrelation (link) between the monitoring devices. A more detailed hierarchical structure can be determined with more monitoring points available for analysis.

A benefit of the machine learning hierarchy algorithm is to automatically provide a basic hierarchical structure of a supply system that is being monitored with minimal user input or none. The hierarchy can be used as a tool for evaluation by other systems 114. Another advantage is that the accuracy of time synchronization between the monitoring devices and the monitoring system software is improved.

In an example in which the supply system 102 is an energy monitoring system, samples of specific electrical parameters (such as power, voltage, current, or the like) are taken simultaneously from each monitoring device in the energy monitoring system. Energy. This parameter data is stored and analyzed with respect to the moment at which the sample is taken, the associated value of the data point, and the monitoring device that provides the data.

The data taken from each monitoring device in the energy monitoring system is compared to each other to determine if there is any correlation between the monitoring devices. Data are analyzed based on trends and statistical correlations, as well as similarities and differences over a predetermined period of time.

According to one example, one or more rules or assumptions are used to determine the hierarchical order of the electrical system. Certain assumptions about the supply system may have to be made in order to automatically learn the hierarchy of the supply system. The assumptions are based on Ohm's Law, energy conservation, and work experience with typical energy distribution and monitoring systems.

The general rules that the machine learning hierarchy algorithm can make in relation to energy systems and energy monitoring systems may or may not include any combination of the following aspects:

1. The power system being analyzed is in a single configuration 320 (figure ⁷⁾ or in a multiple radial feed configuration 330 (figure 8).

2. It is assumed that the meter that measures the highest energy use is at the top of the hierarchical structure (for example, the Main 322 shown in Figure 7), taking into account the inaccuracies in the meters.

3. The sampling data rate by the meters is at least greater than the shortest duty cycle of any load.

4. Energy is not consumed or generated alternatively in the electrical system during the process of collecting parameter data.

5. The error due to the time lag in all the meters of the energy monitoring system is minimal when the data is supplied from the monitoring device to the monitoring system software running on the computing device 132.

Any combination of the following additional parameters may or may not be present for the machine learning hierarchy algorithm:

1. Data are not collected for hierarchical purposes from two monitoring devices installed at the same point of an electrical system, although this parameter is not necessarily a requirement in all aspects disclosed in this document.

2. The no-load meters are ignored or use only voltage, other signal measurements, and / or configuration information to determine their position in the hierarchy.

3. There may be multiple principals in the power system (Main1, Main2, Main3, etc.)

4. The data is provided to the monitoring system software for each monitoring device in the system.

5 Loads that start or stop affect the load profiles for any corresponding data measured upstream with a direct or indirect link with that load.

6. The voltage characteristics (fundamental, harmonic and symmetric components) are relatively consistent for all monitoring devices on the same bus.

7. Transformer losses in the electrical system are minimal with respect to the downstream loads from the transformer.

8. The general correlation (over time) of loads between monitoring devices indicates a direct or indirect link.

9. Multiple charges not measured at one point in the electrical system are added in a single unknown charge.

Any of the above assumptions and parameters can be combined for a radially powered power supply system. For example, any combination of the following assumptions and rules based on rules may or may not be used:

1. The voltages and currents are higher the higher (closer to the top of the hierarchy) is a monitoring device, assuming that upstream transformers or other energy conversion elements are not involved.

2. Harmonic values are generally lower the higher a monitoring device is located.

3. Transformers can vary voltages and currents.

4. The total energy flow is higher in the upward direction than in the downward direction.

5. The feeding system can be a radial feeding system.

6. Two monitoring devices will not be installed at the same point, although this parameter is not necessarily a requirement in other aspects.

7. Monitoring devices with the same voltage distortion are connected adjacently.

8. The total load measured at a specific hierarchical level is equal (excluding losses) to the sum of all measured and unmeasured loads directly linked to that hierarchical level.

Monitoring devices are considered to be at the same hierarchical level if all are linked directly to the same reference device. For example, in reference to Figure 7, a simplified single line diagram of a supply monitoring system 320 with five different levels represented by 323a, b, c, d, e is shown. In the specific case of an energy monitoring system, each level represents a feeder to which multiple monitoring devices can be directly linked. All control devices linked directly to a feeder are considered to be at the same level of feeders. Therefore, the main 322 is directly linked to the feeder 323a and, therefore, exists at its own level of the hierarchy. The feeder 323b is directly linked to three monitoring devices and, therefore, comprises a different level. Feeder 323c comprises another level than feeders 323a and 323b because control devices directly linked to feeder 323c are not directly linked to feeders 323a or 323b. In the case of water, air, gas and steam systems, each level can be represented by a header instead of a feeder.

A specific aspect of the machine learning hierarchy algorithm 400 is presented in the flowchart in Figures 9-11A. Algorithm 400 first checks if there is more than one monitoring device in the system (402), and if not, the algorithm ends. If more than one monitoring device is present, electrical data is taken from each monitoring device (M ¹ , M ² , ..., Mk) and compiled in a Data Table (404). The Data Table tabulates the raw data (such as power, voltage magnitude, voltage distortion, current magnitude, current distortion or symmetric component data) taken at regular intervals (T ¹ , T ² , ...., Tn) for a certain period of time. The period of time between samples depends on the shortest duty cycle of any load in the energy monitoring system. The maximum period of time (Tn) is determined based on the level of variation of the load of each monitoring device in the energy monitoring system. The monitoring device with the maximum power in the Data Table is assumed to be a Principal (that is, the highest level in the electrical hierarchy) (408). However, multiple hierarchies are also contemplated (ie, multiple Main). Below is an example of the Data Table in Table 1.

Table 1: Sample Data Table

i did did did did did k

Once the data in the Data Table has been accumulated, a Verification Matrix is developed. The Verification Matrix is an array of logical connections based on the Data Table. A zero (0) indicates that there is no direct link between any two monitoring devices, and one (1) indicates that there is a possible link between two monitoring devices. An example of a Verification Matrix is illustrated in Table 2 below. In Table 2, it is assumed that there is no link between Meter 1 and Meter 2. This is because the power measured by Meter 1 exceeds that of Meter 2 at an entry in the Data Table and the power measured by Meter 2 it exceeds that of Meter 1 in another entry in the Data Table. Meter 1 always correlates with itself, so an NA is placed in that cell of the Verification Matrix. Only half of the Verification Matrix is required due to information redundancy.

Table 2: Eem of the Verification Matrix

dos:Once the Verification Matrix is determined, the data from each monitoring device in the Data Table is used to develop a Correlation Coefficients Matrix (CCM) shown below in Table 3. In the Coefficients Matrix of Correlation, a statistical evaluation is performed to determine the linear relationship of each monitoring device in the electrical system with respect to the other monitoring devices in the matrix. The correlation coefficient between any two monitoring devices is determined and placed in the appropriate cell in the Correlation Coefficients Matrix. In the example in the following Table 3, C ¹² is the correlation coefficient of Meter 1 with respect to Meter 2. The higher the value of the correlation coefficient, the greater the probability that these two monitoring devices are directly or indirectly linked. On the contrary, the lower this number, the lower the probability that these two monitoring devices are directly or indirectly linked. The following Equation 2 is used to determine the correlation coefficient between any two monitoring devices determined

two:

in which: ^{px, y} is the correlation coefficient and is in the range of -1 <px, and <1; ^{Cov (x, y)} is the covariance of x and y; and ^Ox and ^Oy are the standard deviations of x and y, respectively.

in which: ⁿ is the number of data elements in xe ^y, and ^jx and ^Jy are the average values of xe ^y respectively.

The diagonal cells of the Correlation Matrix are always 1 because each meter has a 100% correlation with itself. Again, only half of the Correlation Matrix is required due to data redundancy (for example, C ¹² = C ²¹ ).

Table 3: Example of correlation coefficient matrix (MCC)

C C C3 C

Returning to Figure 9, a list of meters is developed for each level of the hierarchy under consideration. It is assumed that the highest level is the meter with the highest power reading, which is supposed to be a principal. Once that meter is in the Data Table (408), algorithm 400 places the main one in a list of hierarchy power levels and deletes the list of monitoring devices at the current feeder level of the hierarchy ( 410). In subsequent iterations through the MAIN LOOP, algorithm 400 places the reference meter in the previous list of hierarchy feed levels. It should be understood that in the first iteration there is no previous list of levels. Algorithm 400 deletes a Correlation Reference Matrix (CRA) (412), and designates the principal as a reference monitoring device (414). The following Table shows an example of a Correlation Reference Matrix for n iterations for a given feeder level. C ⁵¹ corresponds to the correlation coefficient between meter 5 (the reference meter) and meter 1, C ⁵² corresponds to the correlation coefficient between meter 5 and meter 2, and so on. Initially, the Correlation Reference Matrix is cleaned for each feeder level, and algorithm 400 develops a new Correlation Reference Matrix for each feeder level by filling each iteration column with correlation coefficients for all meters at the level of current feeders. A specific example in relation to Table 5 is explained below.

The Correlation Coefficients Matrix (CCM) is calculated based on the power data (416). In the first iteration, the only known element in the hierarchy is the main one, and the hierarchy is machine learning from the highest level down, according to some or all of the assumptions or parameters listed above.

Table 4: Example of Correlation Reference Matrix (CRA)

Continuing with Figure 10, algorithm 400 zeroes the correlation coefficients in the Correlation Coefficients Matrix for the meters that have zeros in the Verification Matrix and the meters that have already been detected to be connected (418). The column for the reference monitoring device is copied from the Correlation Coefficients Matrix to the Correlation Reference Matrix (420). A specific example will be explained below in relation to the following Table 5. Assume that meter 5 in the Correlation Coefficients Matrix is designated as the reference meter (414). Algorithm 400 calculates the Correlation Coefficients Matrix based on the Data Table (416) and zeroes the correlation coefficient (s) in the Correlation Coefficients Matrix for meters that have zero in the Matrix Verification and the meters that have been detected to be connected (418). The column of the Correlation Coefficients Matrix corresponding to meter 5 is copied in column Iteration 1 of the Correlation Reference Matrix. Referring to Table 5, the meter 11 has the highest correlation with the meter 5 which is equal to 0.649, and the meter 11 is marked as connected with the meter 5 for the current feeder level.

In Iteration 2, the power of meter 11 is subtracted from the power of meter 5 in the Data Table, and the correlation coefficient of meters 5-11 falls to -0.048 in Iteration 2, which provides a high degree The fact that the meter 11 is interrelated with the meter 5. It should also be noted that the correlation coefficients of some meters tend to increase as the iterations progress. For example, the correlation coefficients for the meter 18 with respect to the meter 5 gradually increase from 0.296 in the iteration 1 to 0.417 in iteration 2 to 0.436 in iteration 3 to 0.525 in iteration 4 and finally to 0.671 in iteration 5, which is the highest correlation coefficient among all meters (meter 5 correlated with itself is always 1.0, so its correlation coefficient is ignored). This growing trend also provides a high degree of confidence that meter 18 is also directly linked to meter 5, and this link is finally confirmed in Iteration 5. The same increasing trends are observed, for example, in meters 12 and 15. In iteration 7, none of the correlation coefficients exceeds a threshold, and algorithm 400 proceeds to analyze the next level of feeders. Through Iteration 7, algorithm 400 has determined that meters 11, 12, 14, 15, 18 and 20 are directly linked to meter 5.

Table 5: Example of Correlation Reference Matrix with example correlation coefficients I ió I ió I ió I ió I ió I ió I ió

5 0.060 0.07 0.0 8 0.080 0.0 3 0.0 0 0.005

Referring still to Figure 10, algorithm 400 detects the monitoring device (feeder) in the Correlation Reference Matrix that has the highest correlation with the reference monitoring device (422). If the correlation does not exceed a threshold (0.5 in a preferred aspect), algorithm 400 proceeds to Figure 11A (OP3), as in the case of Iteration 7 in Table 5 shown above.

Otherwise, algorithm 400 determines whether the current iteration is the first iteration for the reference monitoring device (426) and, if not, determines whether the feeder correlation tends to be greater (428). If the feeder correlation does not tend to be greater, algorithm 400 proceeds to Figure 11A (OP3). A major trend is an indication that the monitoring device is probably at the current level of the hierarchy under consideration.

If the current iteration is the first iteration for the reference monitoring device, the feeder is added to the list of monitoring devices at the current level of the hierarchy (430), and algorithm 400 proceeds to Figure 11A (OP2) . The reference monitoring device and the feeder are designated as linked (or interrelated) directly in a connection table (446), and the power associated with the feeder to the reference monitoring device is subtracted in the Data Table (448) . The connection table maintains a list of devices and their interrelationships (for example, if they are directly linked). By subtracting the power from the feeder associated with the highest correlation coefficient with respect to the reference monitoring device, other feeders (monitoring devices) connected to the reference monitoring device will see their correlation coefficients increase. Algorithm 400 returns to the POWER LOOP in Figure 9, and the next iteration continues with the remaining monitoring devices. Turning now to the OP3 function, algorithm 400 determines if all monitoring devices of the previous level have been analyzed (432), and if not, the following monitoring device (feeder) is obtained at the previous level, and the algorithm 400 returns to the POWER SUPPLY LOOP in Figure 9. If all monitoring devices of the previous level have been analyzed, algorithm 400 checks if a connection has been found for all monitoring devices in the hierarchy (434). If yes, algorithm 400 ends. If not, algorithm 400 checks if the highest correlation coefficient in the Matrix of Correlation coefficients exceed a threshold (436). If not, algorithm 400 ends. If so, algorithm 400 determines if more monitoring devices are found for the current level (438). If not, algorithm 400 returns to the MAIN LOOP in Figure 9. If so, the algorithm moves the monitoring devices from the current level to the previous level (440) and empties the Correlation Reference Matrix (442). The algorithm returns to the POWER SUPPLY LOOP in Figure 9 to determine the relationships between the remaining monitoring devices at the current level.

A modified machine learning hierarchy algorithm 500 is illustrated in Figure 11B. Algorithm 500 begins receiving from each energy monitoring device a criterion associated with each monitoring device (502). The criteria can be power, voltage, current, current distortion, voltage or energy distortion. The criterion can also be a parameter associated with any WAGES utility, such as volumetric flow rates, mass flow rates, volumetric flow and mass flow. For example, when the criterion is a voltage distortion, monitoring devices at the same level of the hierarchy will have approximately the same voltage distortion. Additionally or alternatively, the algorithm can use harmonic distortion values to verify the hierarchy determined by the correlations based on power criteria. Harmonic distortion can also be used by the algorithm to better predict unknown candidates with greater precision. For example, a monitoring device may be marginally correlated with a reference device such that the algorithm cannot determine whether or not there is a direct link. Harmonic distortion can exclude or rule out a possible interrelation depending on the harmonic distortion values of neighboring devices at the same level as the monitoring device in question. For example, a different harmonic distortion returned for the monitoring device in question could rule out that it is directly linked to a device of the previous level.

Algorithm 500 calculates a correlation coefficient between a reference monitoring device and any other monitoring device to be interrelated in the hierarchy (504). Algorithm 500 determines the highest correlation coefficient (506) and interrelates the monitoring device associated with the highest correlation coefficient and the reference monitoring device (508). Algorithm 500 checks if there are more monitoring devices to be interrelated (510) and, otherwise, algorithm 500 ends. If yes, algorithm 500 checks whether to use the same reference monitoring device (512) and, if so, recalculates the correlation coefficients (504). Otherwise, algorithm 500 selects a new reference monitoring device (514) and recalculates the correlation coefficients (504).

Another automatic learning hierarchy algorithm 550 is illustrated in Figure 11C. Algorithm 550 begins by receiving electrical parameter data from each monitoring device at periodic time intervals (552). Algorithm 550 places the electrical parameter data in a Data Table that tabulates the parameter data in each time interval (554). A correlation matrix is formed that includes correlation coefficients between combination pairs of monitoring devices (556). Algorithm 550 identifies an interrelation between a combination pair (558) and removes from the Data Table the power associated with the monitoring device for which an interrelation (560) has been identified. If there are no more interrelations to identify (562), algorithm 550 ends. Otherwise, it recalculates correlation coefficients between the remaining combination pairs (564) and identifies another interrelation between the remaining combination pairs (558). This process is repeated until all the interrelations between the monitoring devices have been identified.

Machine learning hierarchy algorithms are operable both in radially fed systems and in radially multiple fed systems. In radially multiple fed systems, the algorithm first determines the main meter that has the highest power, then determines the hierarchy for that system before proceeding to the next system (s) that has the lowest (n) is) power (s). The machine learning hierarchy can also be useful in non-radial feeding systems.

The machine learning hierarchy algorithm has been discussed in several examples in which the hierarchy is developed from the highest level to the lowest. In an alternative example, a machine learning hierarchy algorithm develops a hierarchy from the lowest level based on local events to each level. For example, monitoring devices near an event will 'see' an event, such as turning a load on or off, before it is seen by remote monitoring devices with respect to the event. The algorithm recognizes interrelationships between monitoring devices based on the occurrences of events and the timestamps associated with each monitoring device on when an event was known. By drawing a chronology of when each monitoring device in the system perceives an event, conclusions can be drawn automatically based on the order of time in which the monitoring device perceived that event about which meters are interrelated (directly linked).

Returning to Figure 1, the integrated and automated data monitoring system 100 produces contextual data 108 from the data alignment system 104 and the hierarchical classification system 106. The Contextual data 108 contains the data from each monitoring device in context with any other monitoring device and, therefore, are more valuable to the user. A contextual analysis of the measured data can be performed, which implies an evaluation of the data in a way that aligns or discloses specific external parameters of each monitoring device. The main external parameters of interest include:

The temporary position of the data of each monitoring device in the delivery system 102 with respect to the data of each other monitoring device in the delivery system 102; Y

The spatial position of each monitoring device M in the delivery system 102 with respect to each other monitoring device M in the delivery system 102.

Evaluating all the accumulated monitoring data of the delivery system 102 in context will provide a degree of knowledge about the delivery system 102 that until now was not available. Because the information of the entire system (software and monitoring devices) is integrated through a uniform context, this approach to monitoring a supply system is called Integrated Monitoring (IM). A useful analogy of the Integrated Monitoring approach is the central nervous system of the human body. The brain (software) knows what is happening with the whole body (monitoring devices) with respect to time and position. If a toe receives a blow, the brain sends a signal for the body to react in some way. Similarly, if an electrical event occurs, the Integrated Monitoring algorithms executed by the monitoring system software provide useful information to the user about the symptoms throughout the monitored system, potential sources of the problem, and possible solutions or recommendations.

The integration of data based on the analysis of the data from each monitoring point uses special algorithms (for example, a data alignment algorithm and a machine learning hierarchy algorithm) in the monitoring system software. In the data alignment system 104, subtle but measurable changes in the frequency and amplitude of data from all data sources are analyzed. These changes are used to establish both the common point of data alignment for all data sources and the position of a data source in the electrical system with respect to other data sources. Because the system data integration process is performed automatically in algorithms of the monitoring system software, much of the effort and expense required by the user is eliminated. The more arbitrary and substantial variations of the analyzed parameters offer a faster integration of the system data. There are several benefits associated with Integrated Monitoring that are beyond what is currently available, including:

The Automated Integrated Monitoring approach greatly reduces existing requirements for the user to manually provide detailed information on the arrangement of the power system in order to put the system data in context. The Integrated Monitoring algorithms analyze the data from each monitoring point in the electrical system to automatically determine the system layout with little or no user participation, saving the user time and resources.

The automated Integrated Monitoring approach eliminates the need for special hardware, additional data lines and, in some cases, monitoring accessories. The Integrated Monitoring algorithms analyze the data from each monitoring point in the electrical system to automatically determine the temporal alignment of the system data, saving on user equipment and labor costs.

The automated Integrated Monitoring approach allows for easier configuration of monitoring hardware and software. This is because the Integrated Monitoring algorithms automatically contextualize the monitoring information throughout the system. Once the monitoring devices are in context, the Integrated Monitoring algorithms can automatically make additional decisions regarding hardware and software configuration. An example would be to set the sub-voltage threshold of a monitoring device based on the location of the monitoring device within the electrical system. Once again, the automated Integrated Monitoring approach saves the user time and resources.

An automated Integrated Monitoring algorithm 600 is illustrated in Figure 12. Algorithm 600 begins by sending a command to the monitoring devices to collect frequency data (602). The data coming from the monitoring devices are uploaded or uploaded to the central computing device (604) and the data coming from all the monitoring devices are aligned (606). When all data is aligned, algorithm 600 determines whether the provision of the power system has been completed (610). If so, algorithm 600 ends, and contextual data can be used in other software applications.

If the power system layout has not been completed, algorithm 600 sends a command to the monitoring devices to collect power data (612). The central computing device that runs the Algorithm 600 loads or uploads the power data from the monitoring devices (614) and determines the arrangement of the energy system (616) in accordance with the present invention. This procedure is repeated until the provision of the power system (618) is completed, at which point the algorithm ends.

There are many considerations to take into account when discussing nuances of practical energy monitoring systems. These considerations multiply when sophisticated algorithms are used to perform complex analyzes and provide firm conclusions to the end user.

Basic factors such as the accuracy, number, location, type and monitored load of discrete devices can significantly influence the results of expert-based systems and their subsequent conclusions. Due to these (and other) unknowns and the inability to predict their occurrence, it is desirable to use an algorithm that can operate successfully despite the dynamic attributes of an atypical energy monitoring system.

Complex energy monitoring systems can include multiple devices at a single point, inaccuracies of devices, erroneous configurations and / or various device capabilities. For these complex systems, it is desirable to use improved algorithms that provide one or more of the following improvements:

• Increased efficiency by segmenting large systems into more manageable sections for processing.

• Location and reconciliation of redundant devices installed in a single point of the energy monitoring system.

• Manage the inherent precision dynamics that occur in all real-world devices.

• Error reduction through continuous reevaluations.

• Determine the location and characteristics of various devices of the power system (for example, transformers, etc.).

• Inter-sample interval adjustment of the algorithm using changes in operational parameters.

• Minimize the total samples required using "sampling and evaluation" techniques.

• Establishment of small virtual loads against losses / inaccuracies.

• Identify and suggest the placement of orphan devices for users.

• Indicate possible configuration problems.

These improvements provide benefits such as fast and robust algorithms and accurate results for complex energy monitoring systems. They also provide solutions to problems related to virtual loads and orphaned devices, as well as with configuration problems that could affect the algorithm.

Due to the inherent requirement of the hierarchy algorithm for synchronous sampling of data from all devices, it can be difficult to spatially align extensive energy monitoring systems (both in the number of devices and in their wide distribution). One solution to the difficulty of aligning extensive monitoring systems is to first segment the system into smaller, more manageable parts (as shown in Figure 13), determine the hierarchy of each individual segment, and then compile the hierarchy of segments in the hierarchy. final space of the entire monitoring system. While an arbitrary grouping of devices into segments can be done, there is a high probability that the devices are not yet capable of spatially aligning, because the arbitrary choice of devices for a given segment may allow unrelated devices and / or devices that they are remote from each other be in the same segment. These conditions may result in insurmountable complications to determine relationships between devices and communication latencies that adversely affect the ability to obtain samples of synchronous data. The logical segmentation of the devices is useful to overcome the problems of an extensive monitoring system.

The devices in an extensive monitoring system can be properly segmented by reviewing the internal (configuration) and external (device readings) parameters of the devices and / or software information. Table 1 shows some examples of internal parameters and how they can be used to indicate a more useful segmentation of monitoring devices:

Table 1: Some internal parameters used for device segmentation Internal parameters Possible indication

Active demand synchronization pulse Device (s) installed at or near the main

Communications (including IP address, Devices located within the same device address, baud rate, geographic area due to limitations on distance from sub-network devices More sophisticated devices located at a higher level of the monitoring system hierarchy

Device (s) installed very close

Device (s) installed very close

Devices installed in it

ubicación geográfica

geographic location

Table 2 shows some general examples of external parameters and how they could be used to indicate more useful segmentation of monitoring devices:

Table 2: Some external parameters used for device segmentation

p)

To show how the method of claim 1 works, Figure 14 illustrates a simple monitoring system with thirteen Mi-M ¹³ monitoring devices, two Ethernet gateways ("EGX's") 700 and 701 and a server 703, using an Ethernet 704 and two sub-networks RS-485705 and 706. At this time of the process no hierarchy of the monitoring system is known.

The review of the internal parameters reveals that there are only three IP addresses for the entire system: one for the M ¹ device connected directly to Ethernet 704, one for the EGX 700 and one for the EGX 701. There are six devices (M ³ , M ⁴ , M ⁵ , M ⁶ , Mg, M ¹⁰ ) that use an IP address on the EGX 700, and six other devices (M ² , M ⁷ , M ⁸ , M ¹¹ , M ¹² , M ¹³ ) that use a second IP address on EGX 701. The review of external parameters (voltages in this case) reveals additional information on the hierarchy of the monitoring system.

As shown in Figure 15, it is possible to develop three logical segments 710, 711 and 712 using information derived from internal and external parameters. Each of the three segments 710-712 in Figure 15 is based on the communications configurations. Within these segments, potential relationships are determined based on measured voltages (for example, phase, discrete harmonic components, average, THD, thd, etc.) in each device. The dashed lines in Figure 15 connect devices potentially located on the same bus. It should be noted that device segmentation does not provide a final solution, but it is useful to reduce the number of monitoring devices to deal with at any given time.

The next stage is to determine the hierarchy of the devices within each of segments 710-712, using the algorithms described above. This stage can result in multiple principals within each segment of multiple devices due to the practical limitations of segmenting an unknown monitoring system. Figure 16 illustrates the results after completing this stage. Note that a bus with a dashed line is still shown to indicate a potential relationship.

Once the hierarchy for the devices within each segment has been determined, the most superior devices of the three segments 710-712 (M ¹ , M ² , M ³ , M ⁴ , M ⁵ , Me, respectively) are identified and grouped into a special subset of all monitoring system devices. This special subset of devices is processed with an algorithm to determine their hierarchical order, as shown in Figure 17. Once the hierarchy of the special subset is determined, its results are compiled with the results of the hierarchy of the segments in the final hierarchy. for the entire monitoring system. Figure 18 illustrates the final hierarchy for the simple monitoring system used in this example. It should be noted that a final reassessment of the entire hierarchy to validate the solution provided. If a device or devices are found to be incorrectly located through this reassessment, the algorithm can reposition that device or devices accordingly.

Using the concepts described above, it is possible to conclusively establish the hierarchy of an extensive monitoring system that could otherwise be undetermined. In addition, it is possible to conclusively determine additional aspects of the monitoring system, as explained below in relation to the "Additional Hierarchy Data". While this example only uses an internal parameter (communications configuration data) and an external parameter (voltage data), other unique internal and external parameters may be used instead of or in conjunction with those specified in this example.

Figure 19 illustrates a flow chart for segmenting extensive monitoring systems into manageable segments for processing. After initializing the system in step 801, the internal and external parameters 802 are reviewed to see if there are indications of potential relationships in step 803. All potential relationships are compiled in a composite table in step 804 and then Logically, all the system devices are segmented in step 805, based on the relationships in the composite table. A default or user configuration 806 can be used to allow manual segmentation improvement or to improve segmentation efficiency.

In step 807, the algorithm determines the hierarchy of the devices within each segment, as described above in relation to Figures 15 and 16. Then, all the main devices of each segment in step 808 are determined, and determines the hierarchy of all the main devices in step 809, as described above in relation to figure 17. Finally, in step 810, the segments in the final hierarchical structure for the entire system are consolidated, as described above in relation to Figure 18. In step 811, the hierarchy of all segments is re-evaluated and then the subroutine ends in step 812.

It is preferable that the algorithm allows all the devices that the end user considers necessary to be located at the same point of a monitoring system and that they are still considered for hierarchical purposes. Multiple devices located at the same point of a monitoring system can be located in series and adjacent to each other at the same point in the hierarchy. There may be limitations to determine the exact sequence of device installations at the same point due to the precision considerations associated with instrument transformers ("IT") and the devices themselves.

Devices that are connected at the same point of a monitoring system have unique characteristics that allow easy identification. Some of these device features include:

1. The statistical correlation of its electrical parameters is almost 100%.

2. The measured values of its electrical parameters are almost identical.

3. The device settings may be identical.

The algorithm recognizes devices with almost identical characteristics and marks them as installed at the same point in the hierarchy of the monitoring system. A more detailed examination is then carried out by means of a pseudo-synchronous sampling of data from these devices with the intention of placing them in sequential logical order. The sequential order is determined based on some combination of the following data:

• Power reading

• Voltage readings

• Current readings

• Correlation values with other linked devices.

By analyzing these parameters and using the basic principles of Ohm's Law, a sequential logical order can be determined for a given set of devices within the imposed constraints of precision.

There are thousands of discrete components that make up each individual device in the monitoring system. Monitoring devices often require instrument transformers (CT and / or VT) to provide adequate currents and voltages. Each device (and its accessories) is inherently susceptible to some degree of inaccuracy. Factors that contribute to the inaccuracies of the monitoring system devices and their accessories include temperature variations, aging components, damaged or defective components, poor design, firmware / defective algorithms, incorrect application of components, incorrect configuration, bad calibration, etc. .

The accuracy of the monitoring system devices can influence the ability of the hierarchy algorithms to efficiently and effectively determine the layout of a monitoring system. Even seemingly trivial inaccuracies have the potential to confuse an algorithm that relies heavily on fundamental laws of "energy conservation". It is imperative that the hierarchy algorithm be robust enough to tolerate some level of intrinsic inaccuracy within the system.

The system described and illustrated in this document is capable of addressing both configuration and intrinsic imprecision considerations, each of them by different means. The algorithm can detect configuration problems, such as incorrect VT / CT relationships, due to discrepancies in the hierarchy data from the monitoring devices. For example, stronger data correlations occur with devices installed adjacent to each other. These data correlations are subject to relative changes in the data, but not to their magnitudes. Incorrect VT / CT relationships affect the magnitude of the data, but not the relative changes. Therefore, if the analysis shows that there is the greatest correlation between two devices that cannot be linked due to discrepancies in the magnitudes of the data, then it is likely that there is a problem with the VT / CT relationships. Note this condition and be taken into account by the end user accordingly.

Another configuration problem that affects accuracy is the use of scale factors. By scaling the data, significant digits can be eliminated through the rounding process, which affects the accuracy of the data. Although the loss of significant digits may not be critical for large load changes, it has the potential to significantly impact small load changes. Because the hierarchy algorithm is based on load variations between monitoring devices, the quantification of data from one or more devices potentially conceals correlative relationships. In order to detect this problem, the algorithm checks the devices to determine if they are configured to scale the relevant data. If scaling is detected, the algorithm can disable scaling during data collection or notify the end user to do the same.

Although there may or may not be configuration problems, the inaccuracies inherent in monitoring devices (and their instrument transformers) are always a dynamic to take into account. As already said, there are many sources that can contribute to this problem. Even if these sources are somehow mitigated, each device is limited in its ability to measure data with some degree of accuracy.

Figure 20 provides a simple illustration of how the trivial inaccuracies of the devices have the propensity to confuse an algorithm based on the principles of "energy conservation." In this figure, three devices M ¹ , M ² and M ³ show the power to have the same directional flow (down). The main device M ¹ that measures the total power in the system shows a power flow of 79kW. The two devices M ² and M ³ in descending direction with respect to the main device M ¹ measure 35kW and 45kW, respectively, for a total of 80kW. It is evident that M ¹ <(M ² + M ³ ), defying the basic laws of conservation of energy. Since this is impossible, there are two logical conclusions that can be drawn from this discrepancy: (1) that 1kW of power is injected into the system through some supplementary (doubtful) route, or (2) that M ¹ devices, M ² and / or M ³ do not provide accurate measurements (plausible). In summary, the combined inaccuracies of these devices (and / or their accessories) will cause the hierarchy algorithm to select device M ³ as the child of device M ¹ (because it has the highest correlation with device M ¹ due to its greater load) and place the M ² device as an orphan device. The end user will have to manually move the M ² device as a child under the M ¹ device, which is not optimal due to the required user interaction.

An algorithm is able to handle these inconsistencies and provide the correct hierarchy to the end user without the need for manual intervention. The algorithm examines a combination of the correlation coefficient and the quantities measured from the time series data received from the monitoring system devices. To use the system in Figure 20 as an example, the time series data from the M ² device can be correlated with the difference between the time series data received from the M ¹ and M ³ devices (M ¹ - M ³ ). The correlation coefficient is approximately the unit, which means unquestionably that the device M ² must be the child of the device M ¹ . The time series data from the M ² device is slightly greater than the difference between the data from the M ¹ and M ³ devices (ie M ² > “M ¹ - M ³ ), so a tolerance is allowed . The M ² device is then correctly positioned as a child under the M ¹ device despite the discrepancy in the power flow.

Each monitoring system must obey the laws of "energy conservation" - the total flow of energy entering a system is equal to the total flow of energy consumed by that system, whether through charges or losses. Simply put, all the parameters that are being measured must be taken into account. Figure 20 illustrates a system in which it appears that more energy is being consumed than is being supplied (1kW). On the contrary, it may seem that other systems are supplying more energy than is being consumed.

All systems experience some degree of losses, inaccuracies and / or unmonitored loads. In order for the system to balance, it is necessary to take these factors into account. It is a difficult task to discriminate which of these factors may be producing a discrepancy in energy consumption and to what extent. In addition, a discrepancy in energy may be due to any combination of these three factors.

To handle these discrepancies, the algorithm is preferably based on a fixed or relative threshold (for example, a percentage) (default or configured by the end user) to determine how to categorize foreign energy consumption. It is also possible to analyze these extraneous energy consumption in the form of some combination of losses, inaccuracies and / or unmonitored loads, depending on the preference of the end user. To illustrate how this is done, Figure 21 shows an electrical system with 36kW of uncounted power consumption. In this case, the algorithm is configured to assume losses and inaccuracies within the electrical system of approximately 1.5% (or 2kW). The residual power is assumed to be an unmonitored load, which results in a virtual load of 34kW. A VM virtual meter reflects that unmonitored load.

Figure 22 shows another electrical system with 2kW of uncounted power. With the same configuration for losses and inaccuracies (approximately 2.1%), the residual power is negligible and the electrical system is considered to have monitoring devices installed in all circuits. Again, the end user can configure the residual power loss as he deems appropriate.

The precise determination of the hierarchy of a monitoring system requires time to collect in a synchronized manner and then process / analyze data from each device. Comparatively speaking, data collection takes much longer than data processing / analysis due to the correlative aspects of the hierarchy algorithm that require load variations to function properly. Due to this fact, it is beneficial to consider the process of data collection to improve the efficiency of the hierarchy algorithm.

Two techniques are used to improve efficiency and speed with which the hierarchy of the monitoring system can be obtained. The first technique focuses on the periodicity of data sampling, and the second on the number of data samples to be taken. Used in concert, these techniques can optimize processing time to solve a hierarchy of monitoring system.

All energy consumers experience changes in the parameters and operational characteristics of their electrical system over a given day due to their routine operation. For example, when each circuit is closed or open, the load of an installation fluctuates accordingly. Because each monitored system is distinguished by its unique operation and / or load types, there is no predetermined number of data points that is optimal for all types of energy consumers. Figure 23 illustrates an example of a load profile of an energy consumer over a period of 24 hours.

By their very nature, correlative statistical functions (similar to the function on which the hierarchy algorithm is based) are more effective and efficient by analyzing data that shows frequent changes. On the contrary, data that does not change or fluctuate slowly is more difficult when evaluated with correlative functions. In order to accommodate the unpredictability of any operational characteristics of a given installation, an "adaptive procedure" can be used to determine the optimum sampling frequency necessary to construct the time series of hierarchy algorithm data in the most effective and efficient way from monitoring devices. One approach would be to incorporate an adaptive procedure to perform a continuous analysis of the power change with respect to time (dP / dt) and adjust the sample interval accordingly. As the rate of change of the load profile of a device or installation changes (increasing or decreasing), more data samples are taken.

Figure 24 illustrates an adaptive sampling procedure using the load profile shown in Figure 23. After each data sample is taken from the monitoring devices, the time series data from one or more devices are analyzed to determine whether or not to increase or decrease the interval between samples. If the rate of change of power increases or decreases with respect to the previous data sample, then the interval between samples is decreased. However, if the rate of change of power is relatively constant, then the interval between samples may remain constant or be increased. Query tables can be used or equations can be developed to determine the increase or decrease in the interval between samples based on changes (relative or absolute) in operating parameters such as power. The use of this adaptive procedure to adjust the interval between samples increases the importance of the data, resulting in a more efficient algorithm that provides faster results. It also minimizes communications network traffic between the devices and the computing device that runs the algorithm.

In order to further increase the efficiency of the hierarchy algorithm, the number of data samples taken to determine the hierarchy of a monitoring system may be reduced where appropriate. Although requiring a fixed number of data samples is a simple approach, it can ultimately provide too many or Too little data. The consequence of sampling too much data is a longer period than is necessary to determine the hierarchy of the monitoring system. In addition, it creates unjustified traffic in the communications network system. The consequence of sampling too little data is even worse: an incorrect solution can be provided to the end user.

In order to address these concerns, a "sampling and evaluation" procedure can be used. Data samples are taken from the monitoring devices and are periodically evaluated by the hierarchy algorithm. Similar to the adaptive procedure described above, the sampling and evaluation procedure is based on changes (although not necessarily correlative) in the hierarchy determined to adjust the periodicity with which the data is evaluated. To find more hierarchical changes from one evaluation to the next, a longer interval between evaluations is needed. As each evaluation shows a decrease in hierarchical changes, the evaluation frequency is increased until a certain number of sequential evaluations provide identical solutions. Once these sequential assessments are in agreement, it is assumed that the hierarchy has been determined.

In order to minimize errors in the determined hierarchy of the monitoring system, the algorithm can reevaluate its conclusions both immediately after a device has been placed and continuously after that. The initial placement of a device in the hierarchy is based on its correlation value, in which the correlation is greater than zero. The initial placement also depends on the power in the device being less than (or greater than within some tolerance) the power of the device under which it is being placed (its father). Tolerance provides flexibility for precision problems when two devices are monitoring the same load. For example, if the tolerance is set to 5%, then the device could be placed under a device that simultaneously consumes 5% less energy. Tolerance also provides flexibility for precision problems when all routes are being monitored downstream (Figure 20).

Once the device is placed within the hierarchy and its time series data is subtracted from the residual time series data of the parent device, the correlation between the device and the father is checked again to ensure a decrease in their correlation ( what is expected). The correlation of the device with the parent device must be greater than a correlation threshold. If not, the new relationship determined between the device and the parent device can be broken in case some other device shows a better relationship. In summary, the correlations of the devices are checked both before and after their placement in the hierarchical structure to ensure their correct placement.

There are cases in which the evaluation may initially conclude that a particular device must be located at a point in the hierarchical structure. However, a later evaluation may determine that the original location of that same device is incorrect and that it actually belongs to another location in the hierarchy. For example, if the correlation of a first device with a second device is greater than the correlation of the first device with its current father, and the first device is no longer related to the second device, then the first device of its current father is removed . The first device is treated as an unassessed device and is completely re-analyzed. Extensive systems can be especially prone to this occurrence given the large number of devices to solve.

Figure 25 illustrates a flow chart for a process that implements some of the techniques described above for determining and evaluating a hierarchy. After starting this process in step 901, the process determines if there is more than one device in the system that is being monitored, in step 902. If the answer is negative, the process ends immediately, at 903. If the answer in step 902 is affirmative, the process finds a "main" device, that is, a device that has the maximum power level in the system being monitored, in step 904. The main device is marked as "Connected" and "Processed", and assigned a zero level in step 905. "Connected" means that the main device has been placed in the hierarchy, and "Processed" means that the children of the main device have been identified and linked to the main device in the hierarchy The level of the main device is always called Level 0, then Level 1 is assigned to the children of the main device, Level 2 is assigned to the grandchildren of the main device, and so on.

In step 906, the process searches for the children of the main device, and then step 907 determines if any of them exist. A negative response in step 907 means that the main device has no children and, therefore, the process jumps to step 908 to determine if there are disconnected devices that have not been linked in the hierarchy. An affirmative answer in stage 907 causes the process to advance to stage 909 in which the level is increased by one, for example, from level 0 to level 1, which is the level for the children of the main device ... step 910 the process selects one of the devices of level 1 and then marks the selected device as "Processed"("WorkedOn") in step 911 to indicate that the children of the selected device (grandchildren of the main device) have been identified and linked in the hierarchy. In step 912, the process searches for the children of the selected device, and then step 913 determines if there are any devices left at this level (for example, level 1). A negative response at step 913 means that there are additional devices at this level, and therefore the process repeats steps 910 through 913 for the next device at level 1. An affirmative answer at step 913 causes the process to proceed to step 914 to determine if there is any device at the next level (for example, Level 2). If the answer is yes, the process returns to step 909 to increase the level by one and then repeat steps 910 through 914.

A negative response in step 914 means that there are no disconnected devices that have not been linked in the hierarchy, and therefore the hierarchy for this particular main device has ended. The process proceeds to step 908 to determine if there are disconnected devices (for example, orphans). If the answer is yes, the process returns to step 904 to try to find another main device and, if so, determine its hierarchy. A negative response in step 908 advances the process to step 915 to determine if any device in the hierarchy has been modified. If the answer is yes, the process searches for children of the device modified in step 916, and repeats this search as long as step 915 continues to give an affirmative answer. A negative response in step 915 ends the process in step 903.

Figure 26 illustrates a flow chart for a process that looks for the children of a reference device. This process can be used, for example, for searches performed in steps 906, 912 and 916 of the process of Figure 25. After starting this process in step 1001, the process initializes "MaxDevice" and "MaxCorrelation" in the step 1002. In step 1003, the process calculates correlation coefficients for all devices with respect to the reference device.

In step 1004, the process selects a "test" device to be evaluated, and then step 1005 determines whether the correlation coefficient for that device is greater than the "MaxCorrelation" value. A negative response in step 1005 indicates that the test device cannot be selected as a potential child of the reference device at this time, because a potential child must have a correlation coefficient greater than the "MaxCorrelation" value. Therefore, a negative response in step 1005 causes the evaluation of the test device to end up skipping to step 1006, which determines whether all the devices have been verified. An affirmative answer in step 1005 indicates that the test device is a potential child of the reference device, and therefore the process advances to step 1007 to determine if the correlation coefficient of the test device is greater than or equal to zero (that is, not negative). If the answer is "no", which indicates that the coefficient has a negative value, the evaluation of the test device will be completed by jumping to step 1006, because the test device cannot be selected as a potential child of the reference device.

A positive response in step 1007 advances the process to step 1008 to determine if the power measured by the test device is greater than the power measured by the reference device. Since the test device is a potential child of the reference device, a negative response is normally expected in step 1008, and said response advances the process to step 1011 to determine if the test device is already connected to another device. If step 1008 produces an affirmative response, the process proceeds to step 1009 to determine if the power measured by the test device is within 5% above the power measured by the reference device, that is, within a 5% tolerance for inaccuracies due to rounding errors and the like. If the answer is yes, the process proceeds to step 1010 to determine if the correlation coefficient of the test device is less than 95%. A negative response in step 1009 or an affirmative answer in step 1010 causes the evaluation of this test device to end.

A negative response in step 1010 advances the process to step 1011 to determine if the test device is already connected to another device. This is the first stage for a reevaluation process carried out by stages 1011-1014. If stage 1011 produces a negative response, a reassessment is not required and the process proceeds to stage 1015. An affirmative response in stage 1011 advances the process to stage 1012 to determine if the test device is an ancestor or a child of the reference device. If the answer is yes, the reassessment is terminated by skipping to step 1006. If the answer is negative, the process proceeds to step 1013 to determine if the correlation coefficient of the test device with respect to the reference device is greater than the correlation coefficient of the test device with respect to the main device to which the test device is already connected. If the answer is negative, the reassessment is completed by skipping to step 1006. If the answer is yes, the process proceeds to step 1014 to mark the test device as "not connected", to update the power of the device's parent test and to mark that parent device as "modified." Next, step 1015 sets the test device as "MaxDevice" and the correlation coefficient of the test device with respect to the reference device as the "MaxCorrelation" value. This completes the evaluation of the test device selected in step 1004.

Step 1006 determines if all devices have been evaluated. If not, the process returns to step 1004 to select another test device and repeat the process described above in relation to steps 1005-1015.

When step 1006 indicates that all devices have been evaluated, the process proceeds to step 1016 which determines whether the current value of "MaxCorrelation" is 100. If so, step 1017 updates the power reading from the reference device and mark the test device as connected to the reference device, and then the process ends in step 1018. If the value of "MaxCorrelation" is not 100, the process proceeds from step 1016 to step 1019 to determine if it has been Found any device. If no device was found, the process ends in step 1018. If a device has been found, the process proceeds to step 1020 to update the residual power of the reference device (subtracting the power of the device found from the residual power of the reference device), and to reconstruct the correlation matrix accordingly. Step 1021 then determines if the correlation coefficient of the test device has decreased, and if the response is negative, step 1022 restores the previous residual power of the reference device, and then the process ends in step 1018.

If step 1021 produces an affirmative response, the process advances to step 1023 to determine if the "MaxCorrelation" value exceeds a predetermined threshold. An affirmative answer advances the process to step 1024 in which the test device is marked as "connected to the reference device", and the process returns to the original initial stage 1002 to search for more children. If step 1023 produces a negative response, the process proceeds to step 1025 to determine if "MaxCorrelation" is greater than the correlation of the test device with all other devices. An affirmative answer advances the process to step 1024, and a negative response advances the process to step 1022 to restore the previous residual power of the reference device, and then the process ends in step 1018.

In many monitoring systems there are devices installed in locations that experience minimal load variation or that have little or no load. For whatever reason, the correlative aspect of the algorithm may not be able to determine a location for that device within the hierarchy of the monitoring system. In these cases, other readings or device configuration parameters can be used to try to locate the device in question within the hierarchical structure. For example, voltage measurements (phase, average, fundamental, harmonic component, total harmonic, imbalance, symmetric components, etc.) can be used to determine the specific bus to which a device is connected. Configuration parameters, such as communications configuration data, VT / CT ratios, etc., can be used. to help locate a device within the hierarchy of the monitoring system. In summary, individual internal or external parameters or combinations thereof can be used to successfully locate a "difficult" device.

Devices that the hierarchy algorithm cannot locate are assigned an "orphan" status and are located as a new system adjacent to the other determined systems. The hierarchy algorithm allows the end user to locate (for example, by "drag and drop", etc.) orphaned devices in their locations installed within the hierarchy, and the hierarchy database is updated. Once an orphan device is located, the spatial context of that device's data can be exploited accordingly.

The evaluation of the data of each device of the monitoring system can provide more useful information for the end user. For example, reading the voltage data of each monitoring device allows the algorithm to determine the location of transformers within the hierarchy of the monitoring system. The algorithm can determine the nominal primary and secondary voltages for a transformer using both device voltage readings and data from a lookup table with standard nominal voltages. This concept can be incorporated internationally using some combination of device voltage readings, system nominal frequency and standard nominal voltage lookup table.

Data from the monitoring devices can also be used to provide more details to the single line diagram of the hierarchy. Configuration information can be used to display communication system hierarchies, VT / CT relationships configured on each device, security settings and status throughout the system, and any other configuration parameters. Real-time readings of the devices can be set to a specific device within a hierarchy console. The data can be viewed from a discrete device, from an advantageous position for the entire system or from any combination thereof. Ultimately, the end user can comfortably access more information in a more useful format due to its spatial context.

Providing a procedure to automatically determine a hierarchy of a monitoring system is an extremely valuable tool. The real value lies more in the context of the monitoring devices within the monitoring system than in a simple representation of the monitoring system (although representation can also be useful). There are many applications that can be derived from the knowledge of where a monitoring device is located with respect to any other device. The goodness of a hierarchy of a monitoring system is that it can be useful even if it is wrong.

To illustrate this point, suppose that a hierarchy of a monitoring system has been determined as shown in Figure 27 for a small system using the hierarchy algorithm. The end user believes that an orphan device M ⁴ should be placed under a parent device M ¹ , and therefore moves M ⁴ down like a child device as shown in Figure 28. However, the algorithm is capable of determine that the relocation of the M ⁴ device below the M ¹ device alters the logical order of the hierarchy because more energy is now consumed through the M ⁴ device than what is being passed through the M ¹ device. By evaluating the correlations between the parent device M ¹ and its supposed children (M ² , M ³ and M ⁴ ), it can be logically determined whether the device M ⁴ is really the child of the device M ¹ . In addition, it is possible to raise the hypothesis of the root cause of the problem by evaluating the configuration settings of each of the potentially related devices (M ¹ , M ² , M ³ , M ⁴ ), and notify the end user accordingly.

For this example, the correlation between the time series data of the M ⁴ device and the residual time series data of the M ¹ device would first be evaluated. In this case, suppose the correlation is extremely high, which means that there is a high probability that the M ¹ and M ⁴ devices are related. The internal and external voltage and current data of the M ⁴ device will then be compared with the M ¹ , M ² and M ³ devices to locate any discrepancies. In this example, the CT ratio in the M ⁴ device seems to be deactivated by a factor of 10. Hypothetically adjusting the CT ratio by a factor of 10, the algorithm determines that there is a high confidence that the placement by the end user of the M orphan device ⁴ is correct. This conclusion can be validated by evaluating the time series data of the M ⁴ device with respect to the residual time series data of the M ¹ device with the adjusted CT ratio. The end user will be notified of the conclusions (for example, the CT relationship has been incorrectly configured) and will be allowed to make his own evaluation if he wishes.

The example in this section is just an illustration of how the algorithm is able to identify and indicate potential configuration problems based on adjustments of the hierarchy of the end user. As shown above, if the end user requests an adjustment to the hierarchy of the monitoring system, the algorithm evaluates these changes to ensure their agreement within the entire hierarchical structure. These evaluations promote the integrity of both the software (including the hierarchy of the particular monitoring system) and the hardware configurations.

While particular embodiments and applications of the present invention have been illustrated and described, it should be understood that the invention is not limited to the precise construction and compositions disclosed herein and that various modifications, changes and variations may be apparent from the above descriptions without departing from the scope of the invention as defined in the appended claims.

Claims (4)

1. A machine learning procedure of how multiple monitoring devices (Mi to M ¹³ ) are directly or indirectly linked in a monitoring system, in which the monitoring system uses an Ethernet (704) and comprises at least one Ethernet gateway (700, 701), the procedure comprising: determining an IP address for each of said monitoring devices (Mi to M ¹³ ), in which the monitoring devices connected to the Ethernet through the same Ethernet gateway (700, 701) have the same IP address; receive voltages measured by each of said monitoring devices; grouping said monitoring devices into multiple segments based on the determined IP addresses, in which the monitoring devices having the same IP address are grouped into one segment; and determine, within the segments, potential relationships between the monitoring devices based on said measured voltages received. 2. The method of claim 1, further comprising receiving at least one parameter selected from the group consisting of power, waveform distortion, sequential components, flashing levels and temperatures measured by said control devices. 3. The method of claim 1 wherein said monitoring system is an electrical energy monitoring system. 4. A system comprising means configured to perform the method of one of claims 1 to 3.